How PLACE voting works (full details)

I’ve been writing various articles about how PLACE voting is the best hope to solve gerrymandering in the US. But I’ve been told on Twitter that, in focusing on the benefits of the method, I’ve neglected to explain clearly and patiently enough how it actually works. I hope this article will change that.

(Note: this article is written primarily for a US audience, so I’ll use that vocabulary; in particular, I’ll talk about electoral “districts”. But most of what I say would apply to Canada or the UK too, if you change that to “ridings” or “constituencies”.)

Context: what PLACE voting can do

Before I explain the mechanics of PLACE voting, let me spend two quick paragraphs setting the scene — one on the problem, and one on how PLACE fits into solving it.

Gerrymandering is when politicians deliberately draw district lines in order to reduce voters’ power. It’s democracy turned on its head: instead of voters choosing their representatives, it’s so-called “representatives” choosing their voters. But as bad as gerrymandering is, it’s only one strand in the tangled knot of problems created by our horrible “First Past the Post” voting method. FPTP reduces politicians’ accountability to ordinary voters, and increases their dependence on donors and on the most-polarized primary voters. It makes third parties irrelevant, shutting new voices out of the political debate. It encourages negative campaigning, occasionally breaks down into undemocratic “spoiled” election outcomes, and fosters gridlock and sabotage within government.

As I’ve argued elsewhere, PLACE voting (which stands for “Proportional, Locally-Accountable, Candidate Endorsement voting”), is a key part of a realistic plan to solve these problems nationwide. Because it can give a proportional result — one which takes equal account of the greatest possible number of voters—using existing districts and machines, PLACE voting could be a remedy imposed by congress on exactly those states with the most-gerrymandered maps. In other words, using a 14th-amendment equal-protection argument, Democrats could impose the system on mostly Republican states. Thus, though in the long term the method has benefits for voters of all kinds and could actually reduce the unearned power of the major parties, in the short term, its potential benefits for one of those parties in particular makes it politically feasible.

Outline: main steps of PLACE voting

How does place work? There are three main parts of the process, and I’ll explain each of them separately.

1. Voting: As a voter, you either choose a candidate running in your district; write in a candidate from another district; or simply vote by party.
2. Tallying and transfers: Once the votes are cast, this is the process to find winners. The weakest candidates are eliminated and votes are transferred until there’s one winner per district. If possible (and it usually is), each winner will a same-sized pile of votes, and the pile of unassigned votes will be smaller than each of those.
3. Extra territories: To ensure that voters have a representative even if somebody they voted against won their district, each district is assigned as “extra territory” to one representative from each winning party except the one that won locally.

Of these three steps, the only one that you really need to understand in order to vote is the first. But of course, in order to convince you to become an excited supporter of PLACE voting, I have to explain all three.

Step 1: voting

Here’s a sample PLACE ballot:

This ballot presents a simple choice of who you vote for — a local candidate, an out-of-district write-in, or a party. As it says on the ballot, if your first choice is eliminated, the votes they’ve received will be counted for another candidate they endorsed (that is, whichever of the listed “similar candidates” got the most direct votes). If you choose “other candidate” but don’t write in a specific choice, your vote will be divided among all the “other candidates” in the party, in proportion to how many direct votes each received; and thereafter, your vote will then transfer as if it had been cast directly (as divided).

Step 2: Tallying and transfers

Once the votes are cast, PLACE has to find the winners. As John Cleese said in his 1980s ad for proportional representation, this step “is the polling officer’s problem,” but I’ll explain the details here anyway. (Cleese was actually talking about STV, single transferable vote, which is an ancestor to PLACE voting, but the principle is the same.)

Relevant disclaimer at 2:31

In order to explain this process fully, I need to cover three things: the basic logic of transferable votes, the rules for deciding where to transfer a vote, and the rules for deciding which candidates to eliminate.

Basic logic of transferable votes

Let’s get philosophical for a moment. What’s the point of electing a legislature? It’s the ideal of a democratic republic: to choose representatives who will deal with the details of legislating in the interests of their constituents. Ideally, then, if each representative is to have the same vote in the legislature, they should each represent the same number of constituents, and do so as faithfully as possible.

In that ideal world, then, each voter’s ballot would count for just one representative. If you want your rep to be as faithful as possible to you, you can’t also be asking for your vote to affect who gets to represent somebody else.

(Of course, a democratic republic also puts some constitutional bounds on what laws the legislature can pass. It’s probably best to avoid having representatives who’d rather subvert that constitution than work with their peers in the interests of the country. So it’s reasonable to have some rules that encourage winners to have at least some amount of broader appeal. I’ll get back to this point later.)

So the goal is to find a given number of winners, such that each winner has an equal pile of votes, and as few votes as possible are not in some winner’s pile. The votes in each candidate’s pile should be as “close” to that candidate as possible; perhaps that candidate was not the voter’s first choice, but they should still be a good representative.

That’s is what the system of transferable votes is designed to do. In order to do that, you need to define how big each pile should be, which we call a “quota” (technically speaking, a Droop quota). The formula is [(number of votes)/(number of seats + 1)] + 1. This guarantees that the leftover pile of unused votes at the end of the process will be almost, but not quite, as big as each winner’s pile.

Here’s a couple examples of this quota for different numbers of votes and seats:

• Say 200 people are electing a club president. That’s a one-seat election, so the quota would be 101 — that’s the familiar 50%+1
• Say a city of 10K voters is electing a 9-member city council. A quota would be 1001 voters, or 10%+1. If the city were divided up into one-seat districts, that quota would be just over 90% of the average district.

Once you’ve defined the quota, the transfer process is easy.

• Tally the votes initially.
• Whenever a candidate reaches the quota, they are counted as elected, and no longer able to receive vote transfers. Any votes they had above the quota are transferred. There are various ways to choose which votes to transfer, but most modern systems transfer a fraction of each of the votes, choosing the fraction to ensure the correct total number of votes are transferred.
• If no candidate reaches the quota, the weakest remaining candidate is eliminated, and their votes are transferred.

That’s it. There’s dozens of videos on YouTube which show this procedure graphically in the case of STV. For instance, if you like Scottish accents, here’s a cute one. PLACE voting is different from STV because of the rules below, but the vote transfers work the same, so the only parts of those videos that don’t apply are the parts about numbering choices on the ballot.

Where is a vote transferred? (Partial, predeclared delegation)

In PLACE voting, voters simply pick one candidate (or party). So how do you know where to transfer those votes if that candidate is eliminated? Through partial, pre-declared delegation.

Before the election, each candidate has the opportunity to officially endorse other candidates. These endorsements are announced publicly and a full list is available in each ballot booth.

When looking at each other candidate, the decision is simply to endorse or not to endorse; there are not various levels of endorsement. So, from the perspective of a given candidate, that divides up the other candidates into 4 groups:

1. Endorsed, same party.
2. Same party but not endorsed.
3. Endorsed, other party.
4. Other party and not endorsed.

When transferring a vote, it goes to a candidate in the first one of these groups that’s still available, except that if only group 4 is available, the vote will be exhausted and cease to be counted. Within a group, it goes to the available candidate with the most votes received directly from voters (highest initial vote tally).

That’s it for explaining how the votes are transferred. But there’s still a few questions to answer about why the system works like this.

• Why not just use STV, and let voters rank all the candidates?

The most obvious answer is: for simplicity. Just a week ago, I voted in an STV election here in Cambridge, MA. Here’s a (blurry) picture of my ballot:

So, in order to decide how to vote, I had to read statements from and/or writeups of most of the 26 candidates in the race, then decide what order I liked all of those candidates. That was a lot of work, and most voters aren’t interested in doing all of that. And even if they were, it’s pretty easy to make a mistake on that ballot above and accidentally invalidate your vote.

But that’s not the only advantage of PLACE over STV. Here are some others:

PLACE voting can be tallied and reported entirely at the precinct level, making outcomes far more transparent, and counts/recounts more secure. That’s because there are only a limited number of possible ways to vote (twice the number of candidates plus twice the number of parties), rather than an exponentially-large number as in STV (more than the factorial of the number of candidates, which can be insanely large).

PLACE voting actually helps minority communities have a stronger voice. If a given candidate represents a minority community, then even if she doesn’t expect to win, her endorsement is valuable. Other candidates will listen to her and, if they want that endorsement, compete to offer her community credible promises of support. Under STV, they have far less reason to pay attention to that community, because possibly winning a few individual votes is not worth as much as winning a bunch of votes which have unified their bargaining power under a single leader.

• Why not let each candidate rank all the other candidates, instead of this “limited delegation” system where they only endorse each other?

This idea of predeclared, fully-delegated STV was actually invented in the 1890s, when it went by the name of the “Gove system”. But the problem with that system is that in practice, voters can’t carefully evaluate the basis for every ranking choice. There’s too much room for corruption and logrolling, making it possible for candidates to become unaccountable to voters by being high enough in enough of the rankings of others.

The limited delegation of PLACE makes it easier to hold candidates accountable. That means that weak candidates will likely get few or no transferred votes, even if they get a lot of endorsements. It also means that if a candidate you might support endorses one you hate, you can ask them why; it’s much harder to ask the question “why did you rank that person 3 when clearly they deserve no higher than 5”.

• Why does “same party, unendorsed” come before “other party, endorsed”?

Because that way, party label really means something. If you want to endorse candidates across multiple parties and have your vote transfer back and forth across those parties, then you can always run as an independent.

• Why have delegation at all? Why not just transfer the votes within party and then stop?

That would be a viable voting method, and actually a pretty good one. (Technically speaking, it would be a biproportional, open list method.) But it would not really have any room for independent candidates. If you want independent candidates to be anywhere near an even footing, you pretty much have to have some form of delegation; and in that case, I think the partial delegation in PLACE is the best answer.

Elimination rules

The final thing I have to explain is the rules for elimination. There are three such rules:

• First, immediately after tallying the votes, any candidate who didn’t get at least 25% of the votes from their local district is eliminated. (Except if no candidate got 25% from a certain district, then the top candidate in that district is not eliminated.)

This serves two purposes. First, it ensures that a candidate’s local district has some say in whether they win.

And second, it ensures that small splinter parties without a local base of support don’t get representatives. Those votes will still be transferred, so that party can still have power in deciding which of the candidates from larger parties will or will not win. But in any one area, only at most 3 parties will get their full proportional share of representatives.

This helps avoid situations like Israel, where no single party has more than 25% of the legislature, so that in order to pass laws it takes a coalition of at least 3 parties and often many more. Even if those tiny parties have a proportional number of seats, that situation lets them exercise disproportional power, as they hold the rest of the coalition hostage to their parochial interests.

Under PLACE voting, on the other hand, there would probably be an “effective number of parties” (ENP) under 4; that is to say, 2–3 larger parties, 1–2 medium ones, and a smattering of independent candidates winning 1 seat each. That’s a pretty good ENP number for encouraging healthy coalitions; large enough to break the unresponsive two-party monopoly, but small enough to discourage hostage-taking by splinter parties.

• Second, if a candidate wins a district, all other candidates in that district are eliminated.

This is necessary to ensure that there’s just one winner per district.

• Third, until all seats are filled, whenever the process would otherwise get stuck, the “weakest” candidate is eliminated by finding the one who’s furthest behind the leader in their district.

In practice, this is usually the same as the STV rule which eliminates the lowest vote total. But in a few unusual circumstances, it stops the last candidate in a district from being eliminated, so that there can still be one winner per district.

That’s it. I hope I’ve done a good job explaining PLACE voting. If you have any questions, please ask them in comments.